Agilent E4428C and E4438C ESG Signal Generators

[KungFuJosh]

I picked up an Agilent E4438C 6GHz signal generator. It's an interesting unit so far. My hope is that I can use it instead of my SMIQ/SML sig gens, but I'm not sure it will cut it or not (is that a crazy idea?).

This has some nice features, including ethernet and a web console, which I like. I especially like that you can do whatever you want to some of the HTML/image files before flashing the firmware. Of course, I should probably remember to flash it back to stock if I decide to sell it instead. 🤣



The web console exists, which is nice, and it's kinda cool that you can save the display.png as a screenshot since anything it shows in the console is actually a PNG with updated info.



I know I'm going to need to do some calibration, and there might need to be some repairs too, but I'm not sure yet. I'll take more detailed internal photos if I go deeper for repairs.

Does anybody have a calibration script for these? The script linked in this thread is no longer available.

I got the service software, but I don't have a TME license. I sent an email to see how much that will cost, but I'd guess it would cost less to send this to a cal lab to get it calibrated.

I'll get into the test results in another post, but here's the odd behavior:
There's an UNLEVEL error on lower frequencies, and it basically outputs noise. 250MHz and above look normal so far. Frequency is good with or without external ref. Pretty damn close with internal ref after calibration, and dead on with external ref.

As I mentioned in the counter thread, I really like the fine adjustment for the ref on this thing (10.000,000,001,4 on the counter).

Thanks,
Josh

[KungFuJosh]

These are the self-test results. 4 tests failed. There was a 5th before, test ID 2314- which the self-diagnostic blamed for the error. 2313 seems more believable, but I don't really know much about these yet.




2302 + 2306:


2313:


2311:


Thanks,
Josh

[squadchannel]

Does anybody have a calibration script for these? The script linked in this thread is no longer available.

https://web.archive.org/web/20241001110311/https://github.com/svenschnelle/caltools/blob/master/calesg.py

 

[Rubo]

You might want to check Shahriar's channel, he has repaired a few of those.
He goes into quite deep with some of the videos on how it works and what can be at fault.

[KungFuJosh]

You might want to check Shahriar's channel, he has repaired a few of those.
He goes into quite deep with some of the videos on how it works and what can be at fault.

That's a great idea, thanks!

I get the feeling with a little work this thing could be great. The only problem is that my SMIQ is already great, and I dunno if I want to invest the time or not. I'll watch some of his videos and see how it looks. If there are any parts that are needed, and they're not 20 years obsolete, that would certainly help.

This thing has AM/FM/Pulse, and my SMIQ doesn't have FM or pulse. I can connect my SML03 as an external pulse, and probably FM modulation too. But then that's two large devices on my rack, that this thing could mostly replace...with lots of assumptions involved.

Does anybody know if the E4438C can do two modulation types at once? I tried AM+FM and it yelled at me, so I'm guessing maybe not. The SML03 can do 3 modulation types at once.

Thanks,
Josh

[KungFuJosh]

This is what output looks like right now. I think all the cal data was lost, so this should be interesting. Signal is ugly, and looks like it has modulation applied, but it doesn't. Except in 141 and 142, AM mod is applied. 143 is the LF output, which is a little noisy, but better than the RF output.

I'll check out Shahriar's video when I have some time.

Thanks,
Josh

[G0HZU]

I've got one of these generators here. If you can get it working, I'd strongly recommend you try and find the real-time baseband generator board for it. This option transforms it into something quite special. You already have the low phase noise and high output power options so it would be worth it to try and add the real-time baseband board. It's also possible to have an internal HDD fitted to store waveforms if you add the baseband option.

If it isn't working below 250 MHz then this points to the BFO/downconversion system it uses below 250 MHz. This will probably be an easy fix.

At some point I'd recommend you change the main bulk capacitor in the PSU, especially if it has be run for many hours. It may have already been changed, but this is a common failure point in the PSU. It's best to replace it before it fails as it may take some switching transistors with it when the capacitance falls and the ESR rises too high.

The other thing to be aware of is that these generators consume a lot of power when running, especially if the baseband board is fitted and running. It can easily exceed 200W with some configurations. So it tends to produce a lot of fan noise once it has been running for a while.

[KungFuJosh]

Are you referring to option 601 or 602? I'm not sure if I have that or not. I have the I/Q ports on the back, which is supposed to be associated with those options.

[G0HZU]

Yes, either that or options 001 or 002 if you have an early E4438C model like mine.

I think you should be able to see the options listed in the Info menu. It looks like one of your screenshots shows this.
On the same screenshot, mine lists the fitted options as: 001 005 403 405 406 408 409 410 415 417 421 422 423 506(6GHz) SP1 UNB and UNJ. Most of these are linked with the real-time baseband board.

[KungFuJosh]

I was perusing the service manual, and it appears all the errors I get in the self test are on the A13 Output board. So, at least I know where to look.

Thanks,
Josh

[G0HZU]

If you do take it apart then can you post up a decent image of the output board please? It might be possible to spot a few good places to look for dead parts.

[KungFuJosh]

If you do take it apart then can you post up a decent image of the output board please? It might be possible to spot a few good places to look for dead parts.

I'll start a repair thread and link it here. I'll post photos there however far along I get.

[Co6aka]

Tangentially, has a option generator been figured out for these generators?

[KungFuJosh]

I'll just throw this here in case anybody needs to recap one of these PSUs.

I inspected the A13 output module, and it generally looks good. Since the issue is awkward voltages, I'll start with recapping the PSU, then measure voltages and go from there. C5 has a fat head, and there's an ugly spot on the board near it, so a little suspicious. 🤷

Funny side note- whomever serviced this last was lazy as shit. They didn't even tighten the SMA connectors on the output path. Doing that didn't fix the output issue, but it certainly cleaned it up.

[G0HZU]

I can't be certain but I think I replaced the 220uF 400V bulk cap with a similar part to the original with Mouser part 661-ETXH401N221MR35S

https://www.mouser.co.uk/ProductDetail/Chemi-Con/ETXH401VSN221MR35S?qs=Odivq8ja527co0rbolgDcA%3D%3D

In my case, this cap had obviously failed as it was distorted and the outer wrapping had detached and the PSU was making a bad smell from this cap. My E4438C is old and has about 60,000 hours run time. Since repairing the PSU it has been reliable but I try not to use it very often because it consumes much more power than a regular sig gen and it makes a lot of fan noise. I have lots of other sig gens here I can use so I only use the E4438C when I need the advanced features it offers.

[KungFuJosh]

I have to double check everything (quantities, values, spacing), but this is my Mouser BOM.

I opted for low ESR, and generally better spec / longer life than what was in there.

I like this thing, but it doesn't seem like it could replace both my SML03 and SMIQ06B, so I don't know if I will keep it. I do like the remote access via ethernet a lot though.

One other thing I like about the E4438C is the Ext inputs 1/2 can be used for modulation, so I should be able to use it with the SML03 inputting more interesting modulations. I'll test it out when I'm done (hopefully) repairing it. I can input pulses to the SMIQ from the SML, but I don't think I can do external modulation with it.

Thanks,
Josh

[KungFuJosh]

Here's a real photo of the A13 output board. I took some with my digital microscope, and a video too. I'll see if they came out well enough to post or not.

Thanks,
Josh

[G0HZU]

It would be nice to see an image of the underside of the output board if you still have access to it.

You can see the ADE-12MH mixer there for the sub 250MHz operation and some of this will be on the underside.

It looks like your 220uF 400V cap choice is only an 85degC part according to Mouser.

[KungFuJosh]

Nice catch, thanks. I filtered the search by temp too, not sure how that got in there.  I updated the BOM with a better choice.

I didn't take the back panel off the A13 board, but if I need to repair it, I will take off the back and take more photos.

I think that fat cap was at least part of the issue; it had a poofy head.

Thanks,
Josh

[KungFuJosh]

Recapping was a success! A huge pain in the ass, but a success! Temporary success. 



I replaced all the electrolytics in the PSU. Probably only needed the fat bastard replaced, but I prefer to get it all done at once.

It clearly needs calibration/adjustment since all of that data was lost at some point, but it looks great compared to where it was, and no more failed self-tests.

ETA: Sonsabitches. It eventually failed again. The temporary win was nice though lol. After letting it run for a while, the waveforms look like a mess again- same as before.


Thanks,
Josh

[KungFuJosh]

All the fails seem pretty similar, so I will need to test the voltages going to the A13 module. I'm guessing they'll be okay, and there will be something annoying on the A13 board itself.

Since it worked at first, I'm assuming there's a bad joint (or component) somewhere that works until it gets hot.

I guess you're going to get the photos of the back of the A13 board after all.

Thanks,
Josh

[Anatoliy]

Hi, KungFuJosh,

I'm wondering if you've had any success resolving the e4438c issue? I also have a device of this type that needs repairing, with the same issue (too low output after warming up and self-test failure). I suspect the issue is with one of the 1gc1-4213 amplifiers on the board with the 6 GHz output. Please check picture attached, red one gives ~+ 3 dB amplification@4,5 GHz, while green one ~+10 dB@4,5 GHz.

[KungFuJosh]

Hi, KungFuJosh,

I'm wondering if you've had any success resolving the e4438c issue? I also have a device of this type that needs repairing, with the same issue (too low output after warming up and self-test failure). I suspect the issue is with one of the 1gc1-4213 amplifiers on the board with the 6 GHz output. Please check picture attached, red one gives ~+ 3 dB amplification@4,5 GHz, while green one ~+10 dB@4,5 GHz.

Hi,

I don't know what those values are supposed to be; I would check the service manual to see if there are any hints.

I've not had a chance to work on this yet, but I do plan on getting to it...this year. Hopefully, I'll have time in the next month or two.

Thanks,
Josh